Katie McGinty: The energy economy’s biggest waste problem is already inside the system

As AI drives data center demand toward 945 TWh by 2030, the fastest new energy source isn't generation — it's the third of electricity quietly wasted.

As the world looks for a way to meet rapidly growing energy demand, a quiet shift is underway. Scale still matters, but it is no longer decisive. The advantage now goes to systems that are faster to deploy, more precise in operation, and adaptive under pressure. 

This change has already transformed parts of the modern economy. Smaller, cheaper and more agile technologies are proving capable of outpacing far more complex, capital-intensive systems built for a previous era. They move faster, adapt more easily, and deliver results sooner. 

Energy is next 

For decades, energy strategy has been defined by one idea: build more. More power plants, more pipelines, more transmission. That logic still holds, as new supply is essential with rising demand. But it is also incomplete. 

Another source of energy capacity is emerging – one that does not require waiting years for permits, financing and construction. It is already embedded into the system itself. It is the energy we produce today, but do not use. 

The energy we are already producing 

It starts with a simple fact: globally, a staggering share of energy never performs useful work. As much as half of all energy input is lost as heat across power generation, industry and buildings. 

 
In industrial systems alone, 20% to 50% of energy is wasted as heat rather than applied productively. This is not a marginal inefficiency. It is a parallel energy system—large, continuous and largely ignored. Even now, more than 3,000 terawatt-hours of usable waste heat remains untapped globally each year. To put that into perspective, that is equivalent to roughly three-quarters of annual U.S. electricity consumption, enough energy to power hundreds of millions of homes for a year if captured and reused. 

That is the scale of the opportunity. Unlike conventional supply, it can be accessed far more quickly. 

AI is accelerating the challenge 

The urgency of this opportunity is becoming clear as energy demand grows. 

Consider the infrastructure behind the AI economy. The International Energy Agency projects that data center electricity demand will more than double by 2030 to around 945 terawatt-hours, equal to about the entire electricity consumption of Japan today. In advanced economies, data centers will drive more than 20% of electricity demand growth this decade. 

The standard conclusion is that only a massive expansion of energy generation can keep pace. But that view overlooks a critical point: a meaningful share of that demand is not intrinsic to computation. It is overhead. 

Today’s data centers still operate with an energy profile shaped by legacy assumptions — especially about cooling. Industry-wide, average power usage effectiveness (PUE) remains around 1.5-1.6, meaning roughly one-third of total energy is wasted on non-compute work. 

As these legacy facilities modernize their IT equipment, and new AI factories are developed, that is not a necessity. It is a design choice. Leading operators have demonstrated compliance with <1.3 annualized PUE targets while eliminating water evaporation from the cooling process. In other words, these data centers can cut non-compute related energy by 50%, while avoiding reliance on community resources such as water. At scale, that saved energy becomes immediately available capacity. 

Put differently, improving how we use energy can deliver a new supply without new generation, and it can do so immediately. 

Unlocking capacity through smarter systems 

This is the essence of the “age of agility” in energy: unlocking capacity faster than it can be built. The same logic applies to thermal energy. Heat is the dominant form of energy in the global economy, accounting for nearly half of final energy consumption, yet it is rarely treated as a resource once generated. 

That is a missed opportunity. 

Technologies like absorption chillers make it possible to use heat—not electricity—for cooling. When integrated into systems such as data centers, they can reduce chiller electricity by 90% while converting waste into a valuable workhorse. 

This is not about marginal gains. It is about fundamentally improving the productivity of energy systems. In Europe, for example, waste heat from process industries is roughly equivalent to total EU building heat demand. 

Speed matters 

The advantages are not only technical; they are also economic and operational. Building new power plants, expanding grids, and permitting infrastructure takes years, often decades. Efficiency improvements and thermal integration can be deployed in months. In a system facing rapid growth, that difference in timing becomes a competitive advantage. 

The economics are equally compelling. Capturing and reusing energy reduces operating costs, increases available capacity and improves resilience. Global estimates suggest that waste heat recovery alone could save tens of billions of dollars in annual savings. And heat pump technology has been shown to slash energy bills 32% while cutting emissions by 60% as well.  

None of these eliminate the need for large-scale infrastructure. The world will still need more generation and transmission to meet long-term demand. 

A broader definition of energy supply 

The competitive edge is shifting. The winners in energy will not simply be those who only build the biggest systems. They will be those who combine scale with agility, delivering capacity more quickly, efficiently and intelligently. 

That means treating efficiency and thermal energy not as secondary considerations, but as core elements of energy supply. 

It also means changing how we measure the system. We readily count the energy we produce. We rarely count the energy we lose. In the old model, building bigger was enough. In the emerging one, speed, precision and adaptability will increasingly determine success, and that means efficiency and thermal energy have to be priority resources. 

The energy transition will depend on both large-scale supply and a fast, often invisible layer—efficiency, thermal recovery. Moving fast and smart – as an efficiency first strategy enables – is the winner as compared to the more cumbersome and costly conventional approach. 

The advantage is there for the taking. To lead in this era, focus not only on how much new energy capacity is built, but on how much existing capacity can be unlocked. Prioritize efficiency, unleash thermal recovery and accelerate smarter system design. The future belongs to those who move quickly to capture the energy already generated and just waiting to be used.  

The opinions expressed in Fortune.com commentary pieces are solely the views of their authors and do not necessarily reflect the opinions and beliefs of Fortune.

This story was originally featured on Fortune.com